Skip to content

Resources

The research this lab stands on, read at the source, and the vocabulary of a field that is still naming itself.

What the research says

The work of other people, read at the source. Each entry says what it found and what it can't tell us.

  • Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande · KDD 2024

    GEO: Generative Engine Optimization

    What it found
    The paper that named the field. On GEO-bench, 10,000 queries, citing sources, adding quotations and adding statistics performed best, while keyword stuffing did not perform well. Cite Sources raised visibility by 115.1% for sites ranked fifth and lowered it by 30.3% for the top-ranked site.
    What it can't tell us
    A benchmark result from 2024. Engines have changed since, which is why VIS-001 retests it on live pages.

  • Liu, Zhang and Liang · Findings of EMNLP 2023

    Evaluating Verifiability in Generative Search Engines

    What it found
    An audit of four generative search engines found 51.5% of generated sentences fully supported by citations, and 74.5% of citations supporting their associated sentence.
    What it can't tell us
    The engines audited in 2023 are not today's. The method matters more than the number, and TRU-002 reuses it.

  • Chapekis and Lieb · Pew Research Center, 2025

    Google users are less likely to click on links when an AI summary appears in the results

    What it found
    Browsing data from 900 U.S. adults, covering 68,879 searches. People clicked a traditional result in 8% of visits with an AI summary and 15% without one, and clicked a link inside the summary in 1% of visits.
    What it can't tell us
    U.S. adults in March 2025, and an observed difference. Summaries may appear on searches that draw fewer clicks anyway.

  • Law and Guan · Ahrefs, 2025 and 2026

    AI Overviews Reduce Clicks by 34.5%, and the 2026 update

    What it found
    Across 300,000 keywords, an AI Overview correlated with a 34.5% lower average click-through rate for the top-ranking page. Re-run on December 2025 data, the figure was 58%.
    What it can't tell us
    A correlation from a vendor's own data, on informational keywords. It is useful for direction, not as a law.

  • Google Search Central · Documentation

    AI features and your website

    What it found
    Google's position: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” Traffic from these features is counted in Search Console's Web search type.
    What it can't tell us
    One platform's stated position, and the strongest argument against this lab. Phase IV exists to test it.

The words, as I use them

None of these terms has a settled definition, and people nest them differently. This is how the lab uses them.

AIO
AI Optimization
The umbrella term here: the work of being found, understood and cited by AI systems. Be careful with it, because many people use the same letters for Google's AI Overviews.
AIM
AI Marketing
What AI changes across marketing as a whole, beyond visibility: the funnel, the audience, the content and how results are measured.
GEO
Generative Engine Optimization
The term from the 2024 KDD paper: optimizing content for engines that generate an answer from several sources.
AEO
Answer Engine Optimization
Being the direct answer, in a featured snippet, a voice reply or an AI response.
LLMO
Large Language Model Optimization
Shaping how large language models represent a brand. In my own series I use it for the practical blueprint.